Work

WET: Overcoming Paraphrasing Vulnerabilities in Embeddings-as-a-Service with Linear Transformation Watermarks

Year: 2025

Type: conference-paper

Abstract: Embeddings-as-a-Service (EaaS) is a service offered by large language model (LLM) developers to supply embeddings generated by LLMs.Previous research suggests that EaaS is prone to imitation attacks-attacks that clone the underlying EaaS model by training another model on the queried embeddings.As a result, EaaS watermarks are introduced to protect the intellectual property of EaaS providers.In this paper, we first show that existing EaaS watermarks can be removed by paraphrasing when attackers clone the model.Subsequently, we propose a novel watermarking technique that involves linearly transforming the embeddings, and show that it is empirically and theoretically robust against paraphrasing.1 Training Dataset ` Verification Dataset 0.4 0.6 0 0 0.4 0.6 0.6 0 0.4 0.57 -0.85 1.28 1.28 0.57 -0.85 -0.85 1.28 0.57 Secret Linear TransformationInverse Linear Transformation

Source: Annual Meeting of the Association for Computational Linguistics (ACL)

DOI: 10.18653/v1/2025.acl-long.1122

Authors: Anudeex Shetty, Qiongkai Xu, Jey Han Lau

Institutions: The University of Melbourne, Macquarie University

Language: English

Topic: Adversarial Robustness in Machine Learning

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funders: University of Melbourne, Australian Research Council

Award: AI for Legal Problem Diagnosis in the Diverse Language of Australians